Metals are an important, emerging class of human and rodent carcinogens. Their mechanism of action remains largely unknown. For cadmium (Cd), we have found a clear genetic basis in carcinogenic sensitivity exists. The metal-binding protein, metallothionein (MT), may play a fundamental role in sensitivity to Cd carcinogenicity. The basal and induced activity of the MT gene is clearly associated with tolerance to Cd genotoxicity. Cellular preactivation of the MT gene markedly reduced Cd genotoxicity, while cells with a quiescent MT gene are highly sensitive to Cd genotoxicity. Cd-induced proto-oncogene activation is likewise blocked by prior MT activation. MT is poorly expressed and unresponsive to stimulation in the specific lobe of the rat prostate that is sensitive to Cd tumorigenesis. The testicular MT gene is poorly activated by Cd and does not play a role in the induced tolerance to Cd. Our studies also reveal that Cd is a very effective tumor suppressant. In studies of suppression of nitrosamine-initiated liver tumors in mice by a single iv dose of Cd, given well after tumor formation, it dramatically reduced (greater than 50%) the tumor incidence. Toxicity was tumor-specific, as Cd induced necrosis in the hepatic tumors. MT was not detected in the tumors, even those undergoing Cd-induced necrosis, while the surrounding normal liver cells expressed high levels of MT. Likewise, in human hepatocellular carcinomas, MT was only poorly expressed relative to normal tissue; these might also be sensitive to Cd. Thus, a single, non-toxic dose of Cd dramatically reduces murine liver tumors through tumor cell-specific necrosis, due to down regulation of MT expression and, furthermore, indicates a similar down regulation of MT occurs in human hepatocellular carcinomas. The therapeutic potential of this metal is being further explored. Human MT expression needs investigation in both the etiological and the therapeutic contexts. Lead (Pb) is considered to be the single most hazardous substance in the United States. Exposure of children is of particular concern. Carcinogenic effects of Pb had previously been considered to be mediated through non-genotoxic mechanisms. In a study with mice, we have found that perinatal Pb exposure resulted in renal neoplasia in approximately 50% of offspring. Since these Pb-induced renal tumors developed long after a short-term exposure and in the absence of chronic site-specific toxicity, Pb may be a genotoxic carcinogen, contrary to the prevailing beliefs. This is an extremely important finding in evaluating human risk. Contributions to childhood cancer risk must also be assessed.